Given a list of values e.g. 2015: [27, 30, 47, 40, 45, 27, 27, 24, 22, 39, 31, 25] I need to check the consistency in terms of whether a user is consistent in buying behavior. In a given example above this case is more towards consistency. Another example of worst case would be like 2015: [0,0,0,51,0,0,0,0,0,0,0]. The total amount at the end of year isn't the concern: if a user is consistent in buying behavior then it's labeled as a good case. How can I check this in terms of statistical techniques?

Gini Coefficient is the one which I am looking for.

  • $\begingroup$ Just about every introductory text or course covers measures of spread or dispersion such as standard deviation and interquartile range, which may be what you're looking for. In this case IQR would return 0 for the second example and isn't much use. $\endgroup$
    – Nick Cox
    May 9, 2017 at 8:08
  • $\begingroup$ you should clearly define what counts as consistent for you. For example, is [1, 1, 1, 1, 1, 4, 4, 4, 4, 4] more consistent, less consistent or exactly as consistent as [1, 4, 4, 1, 1, 4, 1, 4, 4, 1, 4] $\endgroup$
    – Glen_b
    May 9, 2017 at 8:48
  • $\begingroup$ If user has bought at least once a month and continue to do so for the rest of the months with the same amount then the user is consistent. 0 for all months doesn't add much information, obviously. In other words consistency of the user is dependent on the previous buying behavior. For example, ideal case is a linear line. If user is moving on that line then user is consistent, distance from the linear line will give us how much user is far from the ideal scenario. $\endgroup$
    – Anonymous
    May 9, 2017 at 8:55


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.